人工智能定量肺结节参数与肺腺癌浸润程度的相关性分析

Correlation analysis of artificial intelligence to quantify lung nodule parameters and the degree of lung adenocarcinoma infiltration

目的探讨人工智能(AI)定量肺结节参数与肺腺癌浸润程度的相关性。方法选取2021年1月—2023年1月医院收治的138例肺腺癌患者(共138个肺结节),根据肺腺癌浸润程度将其分为两组,浸润性腺癌(IAC)归为A组(n=60),微浸润性腺癌(MIA)、原位腺癌(AIS)及非典型腺瘤样增生(AAH)归为B组(n=78)。所有患者均行CT扫描,AI肺结节检测系统分析扫描数据,比较两组AI定量肺结节参数,分析AI定量肺结节参数与肺腺癌浸润程度的关系。结果A组肺结节短径、肺结节长径、最大CT值、最小CT值、平均CT值及恶性概率均高于B组(P<0.05);Logistic回归分析显示,各项AI定量肺结节参数升高均是加重肺腺癌浸润程度的独立危险因素(P<0.05);上述AI定量肺结节参数联合预测绘制ROC曲线分析显示联合检测预测准确率最高,其AUC为0.995;Spearman相关性分析显示,AI定量肺结节参数均与肺腺癌浸润程度呈正相关关系(P<0.05)。结论AI定量肺结节参数对浸润性肺腺癌预测价值较高,且与肺腺癌浸润程度具有一定相关性。

Objective To investigate the correlation between artificial intelligence(AI)quantitative lung nodule parameters and the degree of lung adenocarcinoma infiltration.Methods A total of 138 patients with lung adenocarcinoma(138 pulmonary nodules in total)admitted to the hospital from January 2021 to January 2023 were selected and divided into two groups according to the degree of lung adenocarcinoma infiltration.Invasive adenocarcinoma(IAC)was categorized as group A(n=60),and microinvasive adenocarcinoma(MIA),adenocarcinoma in situ(AIS)and atypical adenomatous hyperplasia(AAH)were categorized as group B(n=78).All patients were scanned by CT,and then the scanned data were analyzed by an AI pulmonary nodule detection system.The parameters of AI quantitative pulmonary nodules were compared between the two groups,and the relationship between AI quantitative pulmonary nodules parameters and the infiltration degree of lung adenocarcinoma was analyzed.Results Short diameter,long diameter,maximum CT value,minimum CT value,mean CT value,and probability of malig-nancy of pulmonary nodules were higher in group A than in group B(P<0.05).Logistic regression analysis showed that the increase of AI quantitative pulmonary nodule parameters was an independent risk factor for aggravating the infiltration degree of lung adenocarcinoma(P<0.05).ROC curve analysis of the above AI quantitative lung nodule parameters combined prediction showed that the combined detection had the highest prediction accuracy,and its AUC was 0.995.Spearman correlation analysis showed that AI quantitative lung nodule parameters were positively correlated with the degree of lung adenocarcinoma infiltration(P<0.05).Conclusion AI quantitative lung nodule param-eters have high predictive value for invasive lung adenocarcinoma and have a certain correlation with the degree of invasion of lung adenocarcinoma.